arXiv subjects

Jean-Baptiste Döderlein

Publications and source records attributed to Jean-Baptiste Döderlein.

3 recordsLinked to original sources

arXiv2026

SpaceTime Programming: Live and Omniscient Exploration of Code and Execution

Programming environments typically separate the world of static code from the dynamic execution of programs. Developers must switch between writing code and observing its execution, often with limited tools to understand the relationship between code changes and runtime behavior. Several paradigms and approaches exist to bridge this gap, including exploratory programming for comparing code variants, live programming for immediate feedback, and omniscient debugging for exploring execution history. However, existing solutions tend to focus on specific aspects and one specific paradigm rather than providing a fully integrated environment with multiple capabilities. This paper introduces \spacetime Programming, a novel approach that unifies these paradigms to create a programming model for exploring both code modifications and execution flow. At the core of our approach is a trace mechanism that captures not only execution state but also the corresponding code changes, enabling developers to explore programs in both space (code variants) and time (execution flow). As a proof of concept, we implemented a Python library supporting SpaceTime Programming and applied it in two contexts: a live omniscient debugger and a Pygame game development tool, showcased through a Flappy Bird-like game. We further evaluated SpaceTimePy on five real-world Python projects, finding performance overhead ranging from 35% to 150% on test suites.

cs.SE↗

arXiv2025

Piloting Copilot, Codex, and StarCoder2: Hot Temperature, Cold Prompts, or Black Magic?

Language models are promising solutions for tackling increasing complex problems. In software engineering, they recently gained attention in code assistants, which generate programs from a natural language task description (prompt). They have the potential to save time and effort but remain poorly understood, limiting their optimal use. In this article, we investigate the impact of input variations on two configurations of a language model, focusing on parameters such as task description, surrounding context, model creativity, and the number of generated solutions. We design specific operators to modify these inputs and apply them to three LLM-based code assistants (Copilot, Codex, StarCoder2) and two benchmarks representing algorithmic problems (HumanEval, LeetCode). Our study examines whether these variations significantly affect program quality and how these effects generalize across models. Our results show that varying input parameters can greatly improve performance, achieving up to 79.27% success in one-shot generation compared to 22.44% for Codex and 31.1% for Copilot in default settings. Actioning this potential in practice is challenging due to the complex interplay in our study - the optimal settings for temperature, prompt, and number of generated solutions vary by problem. Reproducing our study with StarCoder2 confirms these findings, indicating they are not model-specific. We also uncover surprising behaviors (e.g., fully removing the prompt can be effective), revealing model brittleness and areas for improvement.

cs.SE↗

arXiv2024

LiveRec: Prototyping Probes by Framing Debug Protocols

Context: In the first part of his 2012 presentation "Inventing on Principle", Bret Victor gives a demo of a live code editor for Javascript which shows the dynamic history of values of variables in real time. This form of live programming has become known as "probes". Probes provide the programmer with permanent and continuous insight into the dynamic evolution of function or method variables, thus improving feedback and developer experience. Inquiry: Although Victor shows a working prototype of live probes in the context of Javascript, he does not discuss strategies for implementing them. Later work provides an implementation approach, but this requires a programming language to be implemented on top of the GraalVM runtime. In this paper we present **LiveRec**, a generic approach for implementing probes which can be applied in the context of many programming languages, without requiring the modification of compilers or run-time systems. Approach: **LiveRec** is based on reusing existing debug protocols to implement probes. Methods or functions are compiled after every code change and executed inside the debugger. During execution the evolution of all local variables in the current stack frame are recorded and communicated back to the editor or IDE for display to the user. Knowledge: It turns out that mainstream debug protocols are rich enough for implementing live probes. Step-wise execution, code hot swapping, and stack frame inspection provide the right granularity and sufficient information to realize live probes, without modifying compilers or language runtimes. Furthermore, it turns out that the recently proposed Debugger Adapter Protocol (DAP) provides an even more generic approach of implementing live probes, but, in some cases, at the cost of a significant performance penalty. Grounding: We have applied **LiveRec** to implement probes using stack recording natively for Java through the Java Debug Interface (JDI), and through the DAP for Java, Python, C, and Javascript, all requiring just modest amounts of configuration code. We evaluate the run-time performance of all four probes prototypes, decomposed into: compile-after-change, hot swap, single step overhead, and stack recording overhead. Our initial results show that live probes on top of native debug APIs can be performant enough for interactive use. In the case of DAP, however, it highly depends on characteristics of the programming language implementation and its associated debugging infrastructure. Importance: Live programming improves the programmer experience by providing immediate feedback about a program's execution and eliminating disruptive edit-compile-restart sequences. Probes are one way to shorten the programmer feedback loop at the level of functions and methods. Although probes are not new, and have been implemented in (prototype) systems, **LiveRec**'s approach of building live probes on top of existing and generic debug protocols promises a path towards probes for a host of mainstream programming languages, with reasonable effort.

cs.PL↗